gmail-automation

[Skill Scanner] Natural language instruction to download and install from URL detected The document accurately describes Gmail automation capabilities via a third-party managed connector (Rube MCP) and contains no direct code-level malware indicators. The primary security risk is operational: centralizing OAuth tokens, message payloads, and attachments with an external MCP (https://rube.app/mcp) creates a high-value target for credential theft or data exfiltration. Recommendation: treat use as SUSPICIOUS until the MCP operator and Composio's handling of credentials, token lifecycle, data retention, access controls, and auditability are verified; do not connect high-sensitivity accounts without an explicit security review and contractual assurances. LLM verification: This skill's stated purpose (Gmail automation) is consistent with its described capabilities, but it routes all Gmail traffic, OAuth tokens, and attachments through a third-party MCP (https://rube.app/mcp / Composio). That architectural choice creates a significant supply-chain trust risk: the MCP can inspect, store, or modify mailbox data and tokens. The document lacks explicit OAuth scopes, token storage/retention policies, and details on how S3 attachment retrieval is secured. I rate this ski